Metraipolis › archive › dc_ed804eabfe9d

Human-Centric Image Captioning with Subject-Centered Spatial Understanding — arXiv:2609.08300

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08300
License (as archived)
CC0-1.0
Archived
2026-10-16T15:12:10.980Z
SHA-256 of the archived text
639a4a0d4b700c55def754b59b3c1ffec789f80e8458c177658e0b766d4578a6
Stable URL of this page
https://metraipolis.com/documento/dc_ed804eabfe9d

How to cite

Human-Centric Image Captioning with Subject-Centered Spatial Understanding — arXiv:2609.08300 — arXiv (en), https://arxiv.org/abs/2609.08300 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ed804eabfe9d (Metraipolis archive dc_ed804eabfe9d, sha256 639a4a0d4b700c55def754b59b3c1ffec789f80e8458c177658e0b766d4578a6)
BibTeX
@misc{dced804eabfe9d,
  title = {Human-Centric Image Captioning with Subject-Centered Spatial Understanding — arXiv:2609.08300},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08300},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ed804eabfe9d; sha256 639a4a0d4b700c55def754b59b3c1ffec789f80e8458c177658e0b766d4578a6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08300. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

While multimodal large language models (MLLMs) achieve remarkable performance on generic image captioning, they frequently suffer from structural hallucinations in human-centric scenarios. Accurately modeling human subjects is foundational for critical downstream applications, such as accurate avatar/video/image generation and fine-grained human action understanding. However, these tasks require highly precise subject-centered spatial grounding, such as distinguishing egocentric left/right laterality and maintaining correct anatomical-object bindings. Although catastrophic for structural integrity, these localized spatial inversions are often overshadowed by overall descriptive metrics in existing benchmarks. To systematically expose and quantify this bottleneck, we introduce SPACE (Subject-centric Poses, Appearance, and Characteristics Evaluation), a benchmark designed to evaluate subject-centered spatial understanding. On SPACE, we reveal that despite strong generic perception, current MLLMs consistently fail to ground descriptions in the subject's intrinsic frame of reference. To bridge this gap, we propose a specialized data construction and alignment pipeline. We first extract structured spatial hints from fine-grained body-part localization to guide a two-stage caption rewriting process, yielding highly spatially-faithful training data. Furthermore, we design a rubric-based reward for Group Relative Policy Optimization (GRPO) that explicitly penalizes structurally critical spatial errors during alignment. Extensive experiments on SPACE demonstrate our framework significantly improves human-centric caption quality, particularly in subject-centered spatial reasoning, achieving performance competitive with strong closed-source models. Our benchmark and code are available at this https URL .

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.08300 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T14:49:57.300Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).